The present invention is related to method and apparatus for calculating 3D volume of cerebral hemorrhage, in particular to method and apparatus for calculating 3D volume of cerebral hemorrhage based on X-ray CT (computed tomography) images of the head in which the intracerebral hemorrhage has been affected.
In order to diagnose and treat cerebral hemorrhage, 3D volume calculation of cerebral hemorrhage lesion is performed based on the X-ray CT images of the affected head. The 3D volume calculation of cerebral hemorrhage lesion is performed by identifying the location of cerebral hemorrhage on the image, then performing 3D volume calculation of identified lesion. The calculation methods consist of semi-automatic method (see for example patent reference 1) in which a specialist is involved in the segmentation of cerebral hemorrhage lesion, and fully automatic method in which everything is automatically performed from the identification of cerebral hemorrhage lesion to the 3D volume calculation.
[Patent reference 1] JP-A-2005-118510 (pp. 6 to 7, and FIGS. 3 to 8)
In a semi-automatic method, a specialist is involved in the segmentation of cerebral hemorrhage lesion, therefore requiring labor and time. Since the accuracy of the segmentation largely depends on the technical skill of the specialist, the result may be different among specialists and the reproducibility by a single specialist is not sufficient.
In contrast, fully automatic method is economic but has a problem in the segmentation accuracy, and does not yet provide a sufficient result. More specifically, this method has the tendency of under segmentation in the upper cerebral section and the tendency of over segmentation in the middle or lower cerebral section. Moreover, there may be cases that the section that has no cerebral hemorrhage is segmented, and that the cerebral hemorrhage lesion is not sufficiently segmented.
In both methods, the calculation ignores the partial volume effect and the CT value shift among a plurality of images so that the calculation result must be inaccurate.